Re: [Fresco-devel] issue tracking: Request For Comments

On Sun, Jul 28, 2002 at 06:46:05PM -0400, Sean Rae wrote:
> > I'm in the process of setting up an issue tracker for fresco,
> > and I'm designing the entities and their relationships we should
> > use. Therefor I'd like to get your feedback. Here are the types
> > I'd like to use:
> Does this mean you're just in the process of customizing roundup for
> Fresco's use or are you embarking on a new solution?

Customizing Roundup.

> I have a few questions that might or might not be relevant (and many
> apologies if they're not at all relevant):
> 
> - Roundup doesn't seem to mesh with CVS (only one mention of it that I can
> find in the documentation); is there a way to have it keep track of exact
> CVS commits that correlate with issues in this system?  This would be nice
> to find exactly when/where/by whom the bug was handled through the CVS
> tree.

Unfortunately, CVS really sucks to mesh with, not least because
commits are broken up into separate commits for separate directories,
and there's no way to refer to a particular patch as a hole (even if
that patch only affected one directory).  Eventually (possibly in the
next few months), CVS will die, everyone will use Subversion or arch
or something, and this problem will go away.  Until that day, it's
probably not worth the effort.

> - As an aside, wouldn't it be nice to have sample systems doing full
> compiles nightly the way that mozilla does?  Or am I just ignorant of this
> already happening?  It looks as if it would be trivial to write a python
> script to label each with a green/yellow/red light to tell whether a
> certain patch broke the tree or not (or was still compiling, with a
> yellow).

Kinda tricky -- I'm not sure the nice SPI Inc. people (who host our
repo., web site, etc.) would really like us tying up their poor little
Pentium II web-server with Fresco compiles every night.  Also, while
it might be easy to write a little python script that runs make and
says whether it worked, assigning blame for failures is considerably
less trivial.  I don't know that it would be that useful, either -- a
real test suite (which I'm currently working on the infrastructure
for) would be a serious boon, but a simple doIcompileornot?-type test
tends to happen manually, all the time :-).  (It would be more useful
if there was some way to restrict the commit to a sandbox -- so it
didn't "actually commit" until the test passed -- but there isn't
really any good way to do that.)

> - Would it be possible to attach my 'interest' to an issue in the system
> so that whenever anyone does anything to the issue, that I get emailed?
> There might be a way to do this if I listed myself as the developer who
> was working on an issue...?

I believe this is what Roundup's "Nosy" list is for (analogous to
Bugzilla's "CC" list).

> - Would this be a good time to institute a way of keeping track of who
> touched what?  This would be helpful in the future if anyone wanted to
> change the license of the code and wanted approval of everyone who
> actually changed it.  This doesn't seem likely to happen, but is a
> definite what-if scenario.

This is a job for revision control software, not the bug-tracker.
"cvs history" and similar do what you want already.

> Other than my reveries above, your list seems very complete and I can't
> think of anything to add to it, except that I can imagine that bug's
> 'type' attribute will probably need to be expanded from time to time.

I suspect so :-)

-- Nathaniel

-- 
So let us espouse a less contested notion of truth and falsehood, even
if it is philosophically debatable (if we listen to philosophers, we
must debate everything, and there would be no end to the discussion).
  -- Serendipities, Umberto Eco

This email may be read aloud.